High-Dimensional Learning Dynamics of Quantized Models with Straight-Through Estimator

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ichikawa, Yuma, Kashiwamura, Shuhei, Sakata, Ayaka
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915548907962368
author Ichikawa, Yuma
Kashiwamura, Shuhei
Sakata, Ayaka
author_facet Ichikawa, Yuma
Kashiwamura, Shuhei
Sakata, Ayaka
contents Quantized neural network training optimizes a discrete, non-differentiable objective. The straight-through estimator (STE) enables backpropagation through surrogate gradients and is widely used. While previous studies have primarily focused on the properties of surrogate gradients and their convergence, the influence of quantization hyperparameters, such as bit width and quantization range, on learning dynamics remains largely unexplored. We theoretically show that in the high-dimensional limit, STE dynamics converge to a deterministic ordinary differential equation. This reveals that STE training exhibits a plateau followed by a sharp drop in generalization error, with plateau length depending on the quantization range. A fixed-point analysis quantifies the asymptotic deviation from the unquantized linear model. We also extend analytical techniques for stochastic gradient descent to nonlinear transformations of weights and inputs.
format Preprint
id arxiv_https___arxiv_org_abs_2510_10693
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle High-Dimensional Learning Dynamics of Quantized Models with Straight-Through Estimator
Ichikawa, Yuma
Kashiwamura, Shuhei
Sakata, Ayaka
Machine Learning
Disordered Systems and Neural Networks
Artificial Intelligence
Statistics Theory
Quantized neural network training optimizes a discrete, non-differentiable objective. The straight-through estimator (STE) enables backpropagation through surrogate gradients and is widely used. While previous studies have primarily focused on the properties of surrogate gradients and their convergence, the influence of quantization hyperparameters, such as bit width and quantization range, on learning dynamics remains largely unexplored. We theoretically show that in the high-dimensional limit, STE dynamics converge to a deterministic ordinary differential equation. This reveals that STE training exhibits a plateau followed by a sharp drop in generalization error, with plateau length depending on the quantization range. A fixed-point analysis quantifies the asymptotic deviation from the unquantized linear model. We also extend analytical techniques for stochastic gradient descent to nonlinear transformations of weights and inputs.
title High-Dimensional Learning Dynamics of Quantized Models with Straight-Through Estimator
topic Machine Learning
Disordered Systems and Neural Networks
Artificial Intelligence
Statistics Theory
url https://arxiv.org/abs/2510.10693